Building the “See Something, Say Something” AI for Every Camera
[HPP] Martin CasadoDecember 16, 202538 min
26 connections·40 entities in this video→Founding Vision & Personal Motivation
- 💡 The CEO, Shikhar Shrestha, was a victim of an armed robbery at age 12, which sparked his mission to make every camera intelligent after realizing no one was watching a security camera during the incident.
- 🎯 Ambient's core mission is to prevent physical security incidents from happening by transforming existing cameras into smart, proactive monitoring systems.
- 🧠 The initial inspiration came from early image captioning models in 2016, foreseeing AI's ability to understand images and videos better than humans.
AI's Role in Physical Security
- 🔑 AI enables cameras to act as a "see something, say something" node, detecting suspicious precursor behavior before an incident escalates.
- 🚀 Ambient's AI capabilities span prevention, real-time response, and forensics, allowing for immediate action and efficient investigation of past incidents.
- ⚡ The advent of Generative AI and Vision Language Models (VLMs) significantly enhanced AI's reasoning ability, allowing it to understand context and differentiate between benign and suspicious activities.
Ambient's Technology & Approach
- 🔬 Ambient developed its own reasoning VLM called Pulsar, optimized for security camera feeds, making it 50 times more compute-efficient and outperforming public models on this specific problem.
- ✅ The company adopts a privacy-aware approach, focusing on detecting suspicious actions and behaviors rather than using facial recognition or identifying individuals.
- 🛠️ Ambient integrates a human-in-the-loop 24/7 operation to review low-confidence AI alerts, which also serves as a crucial source of training data for continuous model improvement.
Business Evolution & Market Strategy
- 📈 Ambient primarily targets large enterprises such as corporate campuses, hospitals, data centers, and critical infrastructure, where the need for complex security solutions is most magnified.
- 🏡 The product is software that integrates with existing physical infrastructure (cameras, access systems), providing a plug-and-play solution for automation and incident prevention.
- 💬 Navigating challenges like COVID-19 involved repositioning the product to active verticals and accelerated the trend towards cloud-based, remote monitoring solutions in physical security.
Future of Agentic Security
- 🔮 The future of physical security involves AI not just detecting threats, but also performing assessment and automated response, suggesting and executing steps like locking down buildings or calling law enforcement.
- ⏱️ This agentic approach is crucial because seconds matter in preventing incidents, and AI can execute complex workflows more reliably and quickly than human operators in crisis moments.
- 🌐 Ambient's system can connect various data points, from license plate detection to access control events, to provide end-to-end investigation and proactive threat detection across an entire site.
Knowledge graph40 entities · 26 connections
How they connect
An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.
Hover · drag to explore
40 entities
Chapters20 moments
Key Moments
Transcript143 segments
Full Transcript
Topics15 themes
What’s Discussed
Physical SecurityArtificial IntelligenceVision Language Models (VLMs)Generative AIAutomated ResponseIncident PreventionAI ForensicsData PrivacyEdge GPU ServersGo-to-Market StrategyEnterprise SecurityThreat DetectionHuman-in-the-Loop SystemsCorporate CampusesCritical Infrastructure
Smart Objects40 · 26 links
Companies· 3
Concepts· 27
Products· 6
Events· 2
People· 2